Skip to content
Robotics

track_v3_dual-image

a driving policy for F1TENTH mini-racers

2 min readupdated

This trained control policy is built for autonomous mini-robots on the f1tenth platform. It takes simultaneous feeds from two onboard cameras (camera1 and camera2) along with the vehicle's current state, outputting steering and throttle commands to run laps clockwise or counterclockwise. Under the hood is the compact 0.45B-parameter VLA model (Vision-Language-Action — an AI model connecting vision, text commands, and physical actions) lerobot/smolvla_base by Szymon-Ilnicki. Unlike the base model, this version is fine-tuned specifically for dual-stream autonomous driving using the LeRobot framework. It is useful for robotics developers and enthusiasts working with the F1TENTH platform or building custom visual navigation policies on SmolVLA.

Key features

  • Dual-camera vision: processes two 424×240 video streams (camera1 and camera2) alongside a 2-dimensional robot state vector.
  • Bidirectional driving: trained to follow text prompts for running track laps both clockwise and counterclockwise.
  • Compact footprint: the 0.45B-parameter SmolVLA backbone is built to run on consumer hardware without heavy server accelerators.
  • LeRobot integration: fully compatible with the standard lerobot-rollout pipeline for quick deployment on a physical robot.

What you can do

  • Run autonomous laps on a physical f1tenth chassis by hooking up two cameras and specifying the driving direction in a text prompt.
  • Use the checkpoint as a ready-made starting point for fine-tuning on your own track or camera setup via lerobot-train.
  • Explore how multimodal SmolVLA policies behave in practice on visual vehicle-control tasks.

Example

lerobot-rollout \
  --strategy.type=base \
  --robot.type=f1tenth \
  --robot.port= \
  --robot.cameras="{ camera1: {type: opencv, index_or_path: 0, width: 640, height: 480, fps: 30}, camera2: {type: opencv, index_or_path: 1, width: 640, height: 480, fps: 30}}" \
  --policy.path=Szymon-Ilnicki/track_v3_dual-image \
  --task="complete laps around the track anticlockwise" \
  --duration=60
Read moreDetails
Training: 20,000 steps (batch size 8, lr 0.0001, adamw optimizer) on the local/track_v3 dataset of 12 episodes (4,870 frames at 10 FPS).
Framework version: LeRobot 0.7.0, Apache 2.0 license.
Limitations: real-robot test results are not yet listed in the model card; camera names at runtime must strictly match the keys used during training.
Discuss Telegram 0 27

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.1 GB · ready files1.7 GB
  • FP8 / INT8recommendedweights 0.4 GB1.0 GB
  • 4-bitweights 0.2 GB0.8 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # Szymon-Ilnicki/track_v3_dual-image
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download Szymon-Ilnicki/track_v3_dual-image --local-dir ~/models/track_v3_dual-image
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to track_v3_dual-image
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

  1. SmolLM2-360Mfamily root
  2. SmolLM2-360M-Instruct
  3. SmolVLM-500M-Instruct
  4. SmolVLM2-500M-Video-Instruct
  5. smolvla_base
  6. track_v3_dual-image

New posts about versions of SmolLM2-360M (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 8All on Hugging Face
Merges 1All on Hugging Face

About the model

Szymon-Ilnicki/track_v3_dual-image
Author
Szymon-Ilnicki
Category
Robotics
Released
Posted

On NeyroLearn

Hugging Face

270

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments